kegg-pathway-analysis

Pass

Audited by Gen Agent Trust Hub on Jul 10, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSDATA_EXFILTRATIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references and provides instructions for using standard bioinformatics packages, including the R packages clusterProfiler, KEGGREST, and R.utils, as well as the Python package gseapy.
  • [DATA_EXFILTRATION]: The skill performs HTTP GET requests to rest.kegg.jp. This is the official endpoint for the Kyoto Encyclopedia of Genes and Genomes (KEGG) API, used here to retrieve pathway-gene mappings.
  • [COMMAND_EXECUTION]: Provides R and Python code snippets for executing statistical enrichment tests and managing local cache files via json.dump and os.path.exists.
  • [PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes external biological data.
  • Ingestion points: Gene lists and differential expression results (DEGs) provided by the user (SKILL.md).
  • Boundary markers: None present to distinguish data from instructions.
  • Capability inventory: Network access via requests and file system writes for caching (SKILL.md).
  • Sanitization: No sanitization or validation of input gene identifiers is implemented.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 10, 2026, 03:35 PM
Security Audit — agent-trust-hub — kegg-pathway-analysis